add im- and export for IIASA xlsx files

[orichters]

• submission to IIASA database and downloads from IIASA database are in xlsx format. This PR allows to read those files into a quitte object, and to write them such that they can be submitted
• It works, but I'm not so sure whether function names were optimally chosen (I avoided write.xlsx as function name as it already exists in xlsx package), whether tests are sufficient etc.
• I left Michaja as "coauthor" of the function, as I basically copied much of the code from write.mif.
• buildLibrary created lots of changes, hope they are ok…
• improvements or corrections welcome

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

1. Is that desired behaviour? Seems to me putting/leaving NA (or "N/A" or "n_a" or whatever IIASA is using) instead of valid-looking zeros would make more sense …

2. mutate(Value = ifelse(!is.finite(.data$Value) | .data$Value == "", 0, 
                         .data$Value))
   

   is more concise.

[orichters]

Asked the IIASA admins about that. I just adopted Jeromes solution for NGFS, but maybe this is not accurate (anymore?).

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

In any case, an easier (and probably faster, but who is counting) way would be

-        arrange(.data$Period) %>%
-        mutate(!!sym('Value') := ifelse(! is.finite(!!sym('Value')) | !!sym('Value') == "", 0, !!sym('Value'))) %>%
-        pivot_wider(names_from = 'Period', values_from = 'Value')
+        filter(is.finite(.data$Value), '' != .data$Value) %>% 
+        pivot_wider(names_from = 'Period', names_sort = TRUE,
+                    values_from = 'Value', values_fill = 0)

If IIASA says "use 'n_a'", you will have to convert to strings first.

[orichters]

I agreed with IIASA that writing files with empty cells is best. Using NA as values_fill leads to empty cells in the xlsx, so I implemented it like that, no need to convert to string though, luckily.

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

That's quite long-winded … why not pivot_longer(matches("^[0-9]{4}$")?

[orichters]

Well, I copied that from this line…

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

… that actually predates match() (and pivot_longer(), for that matter). :p

[orichters]

@0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q: Should we also add something like this?

            missing.default.columns <- default.columns[! default.columns %in% tolower(colnames(data))]
            if (length(missing.default.columns) > 0) {
                stop(f, " misses the default columns ", paste(missing.default.columns, collapse = ", "))
            }

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

Sure.

[orichters]

Sure.

Opted for warning, I think that is sufficient. as.quitte adds the missing columns with (Missing) and so you can continue with the data.